Overview of ensemble analysis and the chain components (proposals, ...)

[maxhully]

If I'm a new user who hasn't seen ensemble analysis before, I want to get a bird's-eye view of what that is, and each of the decisions that I need to make to set it up---proposal, acceptance, constraints, initial state---with code examples.

[zschutzman]

I think this and a few of the other docs issues (e.g. #295, #294, #279) can be mostly taken care of by conscientiously making tutorials for VRDI.

We should probably take care of any issues that meaningfully affect the way a user interfaces with the software (e.g. #278, #292) before spending a lot of time putting together instructional stuff.

[maxhully]

That's a great point. I'll try and get the ball rolling on some of the API additions/changes that I've been thinking about so that I can get proper feedback from you and everyone.

[drdeford]

I missed this at the time but I will have a couple of (longish) docs addressing exactly these issues ready for eyeballs next week. Right now I have it separated into one piece that just focuses on the theory side - why do we implement Markov chains this way, why check validity rather than computing transition probabilities, what types of proposals exist, etc. - and a separate doc that focuses on the implementation side with lots of specific GerryChain examples/code. There is also a geography data/preprocessing one but that needs to be updated for the new maup code.

I'll send them around next week for feedback.

[maxhully]

Awesome! I tried to orient the maup README around specific preprocessing use cases. I want to update it to use actual data (and maybe make it a notebook?) instead of being purely hypothetical.

[drdeford]

Ok, here are some drafts. The relevant material from the MCMC_Intro is Section 7.
GerryChain_Guide.pdf
MCMC_Intro.pdf

Definitely a work in progress but hopefully this gets us off to a good start. The geography doc that gre out of VRDI last summer here: tinyurl.com/gerrydataguide

[drdeford]

The updated versions that we used for VRDI are here people.csail.mit.edu/ddeford/MCMC_Intro_plus.pdf
and people.csail.mit.edu/ddeford/GerryChain_Guide.pdf as well as the 5 repos from early in vrdi 2019: github.com/vrdi